Neural Text Generators in Enterprise Modeling: Can Chatgpt be Used as Proxy Domain Expert?
Kurt Sandkuhl, Balbir Barn, Souvik Barat
Abstract
Enterprise modeling is concerned with the systematic development of a comprehensive and holistic representation of an enterprise (an enterprise model) to support organisational initiatives. Domain experts have an essential role in enterprise modeling projects (EM), as they provide the required domain knowledge or specifics of the organisation under consideration. The paper investigates if neural text generators (large language models) can reduce the dependency on domain experts for certain tasks in enterprise modeling. The main contributions of this paper are (1) a systematic literature analysis on neural text generator use in EM, (2) the identification of potential for applying large language models in EM, and (3) findings from quasi-experiments comparing output of ChatGPT and domain experts for the same EM task.